Transaction Costs and Market Institutions: Grain Brokers in Ethiopia
Transaction Costs and Market Institutions: Grain Brokers in Ethiopia
Transaction Costs and Market Institutions: Grain Brokers in Ethiopia
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
9<br />
trader’s social capital <strong>and</strong> acts as a positive shifter <strong>in</strong> the trader’s revenue<br />
function. The trader’s revenue maximization is expressed by<br />
(1) = γ L α K β e ε - ωL - νK<br />
where γ is social capital, ω is the opportunity cost of search labor (L), <strong>and</strong> ν is<br />
the opportunity cost of work<strong>in</strong>g capital (K). From the first-order conditions for<br />
profit maximization, the shadow opportunity costs of search labor ω *<br />
<strong>and</strong> of<br />
work<strong>in</strong>g capital ν * are derived as<br />
α<br />
(2) ω<br />
* = R L<br />
β<br />
(3) ν<br />
* = R K<br />
Ord<strong>in</strong>ary least squares estimation of the trader’s revenue function would result <strong>in</strong><br />
asymptotically biased estimators because of the simultaneity bias that exists<br />
because both search labor <strong>and</strong> work<strong>in</strong>g capital depend on revenue <strong>and</strong> thus will<br />
not be <strong>in</strong>dependent of the model’s error term. 5<br />
To overcome this bias, two-stage<br />
least squares estimation is used to obta<strong>in</strong> the coefficients necessary for deriv<strong>in</strong>g<br />
ω* <strong>and</strong> ν*. A rich set of <strong>in</strong>struments for search labor <strong>and</strong> work<strong>in</strong>g capital are<br />
5<br />
This issue generally plagues the estimation of production functions, to which a solution is to<br />
apply duality theory, a solution which fails to use all available <strong>in</strong>formation <strong>and</strong> is statistically<br />
<strong>in</strong>efficient (Mundlak, 1996). In this case, the existence of a rich set of <strong>in</strong>struments provides a<br />
more conv<strong>in</strong>c<strong>in</strong>g <strong>in</strong>strumental variables estimator.